Re: computational model of transactions

paul c a écrit :

> Cimode wrote:
> > Trying to implement decent concurrency on a truly relational system
> > without some abstract model involving network model abstract thinking
> > seems hazardous.
>
> Don't be too sure of that. For example, there is a choice involved in
> interpreting relations as standing for an abstraction of logical
> predicates rather than manifestations of set theory operations. To say
> which comes first is, I think, a philosophical choice that we justify
> one way or another, by some immediate practical reason, eg., whatever IT
> hole we are trying to dig ourselves out of at the time. I imagine this
> might be what's at issue in the first great blunder argument.
Probably but it was not phrazed in a mature way. Order of importance between the two problems is an irrelevant debate. My point is that the issue of concurrency should be dealt with, once RM is dealt with first on the perspective of individual man machine interaction for producing information. Once done, then and only then, the issue of collective interaction with data to produce information can be handled fully. In such matter, I find the concept of logical unit of work, some kind of elementary (atomic) concept too vague and subjective not to add anything but confusion.
> p
Received on Sun Aug 06 2006 - 15:24:18 CEST